Host microRNA interactions with the SARS-CoV-2 viral genome 3’-untranslated region

The 2019 pandemic, caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has marked the spread of a novel human coronavirus. While the viral life cycle is well understood, most of the interactions at the virus-host interface remain elusive. Furthermore, the molecular mechanisms behind disease severity and immune evasion are still largely unknown. Conserved elements of the viral genome such as secondary structures within the 5’- and 3’-untranslated regions (UTRs) serve as attractive targets of interest and could prove crucial in furthering our understanding of virus-host interactions. It has been proposed that microRNA (miR) interactions with viral components could be used by both the virus and host for their own benefit. Analysis of the SARS-CoV-2 viral genome 3’-UTR has revealed the potential for host cellular miR binding sites, providing sites for specific interactions with the virus. In this study, we demonstrate that the SARS-CoV-2 genome 3’-UTR binds the host cellular miRNAs miR-760-3p, miR-34a-5p, and miR-34b-5p, which have been shown to influence translation of interleukin-6 (IL-6), the IL-6 receptor (IL-6R), as well as progranulin (PGRN), respectively, proteins that have roles in the host immune response and inflammatory pathways. Furthermore, recent work suggests the potential of miR-34a-5p and miR-34b-5p to target and inhibit translation of viral proteins. Native gel electrophoresis and steady-state fluorescence spectroscopy were utilized to characterize the binding of these miRs to their predicted sites within the SARS-CoV-2 genome 3’-UTR. Additionally, we investigated 2’-fluoro-D-arabinonucleic acid (FANA) analogs of these miRNAs as competitive binding inhibitors for these miR binding interactions. The mechanisms detailed in this study have the potential to drive the development of antiviral treatments for SARS-CoV-2 infection, and provide a potential molecular basis for cytokine release syndrome and immune evasion which could implicate the host-virus interface.


132
To test the binding of miR-760-3p to the SARS-CoV-2 3'-UTR, we utilized sequences mimicking 133 the respective miR binding site located within the genome 3'-UTR. The T100 and TL sequences (Fig. 1,   134 turquoise, Table 1) were prepared to pre-form a duplex which matches the native fold of the 3'-UTR at the 135 miR-760-3p binding site. To promote formation of this duplex, the T100 and TL samples were brought to 136 100˚C then allowed to slow anneal in a water bath for 1 hour until reaching room temperature (23˚C). Two 137 additional samples, the free T100 and TL sequences, were each boiled for 5 minutes and snap-cooled. 1 138 mM MgCl 2 was added to a fixed 1 μM 3'-UTR T100:TL duplex mimic, along with increasing 139 stoichiometric additions of miR-760-3p (which was previously snap-cooled): 0.5 μM, 1 μM, 1.5 μM, 2 μM. was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted May 19, 2023. ; https://doi.org/10.1101/2023.05.18.541401 doi: bioRxiv preprint for one hour at room temperature (23˚C), and electrophoresed on 15% TBM gels for at 75V and 4˚C for 4 155 hours. The gels were then visualized using SYBR gold stain for identification of all RNA complexes. The 156 experiment was repeated for miR-34b-5p, utilizing the same stoichiometric ratios and procedure.

176
The 3'-UTR T100:TL duplex mimic was prepared using a pyrC tagged Terminus Long sequence 177 (pyrC-TL, pyrC is bolded in Table 1) and T100 sequence at 150 nM in 10 mM cacodylic acid, pH 6.5 which 178 were slow annealed and incubated with 1 mM MgCl 2 as described above for the miR-760-3p PAGE . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted May 19, 2023. ; https://doi.org/10.1101/2023.05.18.541401 doi: bioRxiv preprint experiment. Snap-cooled miR-760-3p was titrated in 12 .5 nM increments to the 3'-UTR T100:TL duplex   180  mimic and equilibrated for 15 minutes prior to recording the emission intensity at 445 nm. Data points were   181  corrected using blank subtraction and miR noise subtraction and normalized to the free 3'-UTR duplex   182 emission intensity. These experiments were repeated in triplicate, and data was fit to equation 1 to determine 183 the dissociation constant K d .

217
To establish specificity of the miR for the SARS-CoV-2 3'-UTR, a negative control was run using

228
The normalized intensity was recorded and fitted to equation 2 for determination of the IC50 value (AAT . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made   255 We first analyzed the binding of miR-760-3p to the SARS-CoV-2 genome 3'-UTR by native 256 PAGE. In the context of the entire genome 3'-UTR, miR-760-3p is predicted to initiate binding at the 257 exposed six-nucleotide bulge located at the 3'-UTR terminus; thus, to mimic this binding site, we pre-258 formed a duplex structure ( Fig. 2A) using two chemically synthesized sequences (Table 1)   a predicted size of 160 nt, as the 3' extension (nt 43-60) of the 3'-UTR T100:TL duplex is predicted to be 271 able to dimerize. To confirm that miR-760-3p is binding to the predicted exposed bulge, we mutated the 272 TL sequence at that site ( Fig. 2A, nt 24-29, bolded in Table 1), which is referred to as the 3'-UTR TL . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

276
To establish the identity of miR-760-3p in the higher molecular complexes from previous native 277 PAGE experiments, we labeled miR-760-3p with a DY547 fluorescent tag at its 5' end (named here DY547- . Further native PAGE experiments were performed to demonstrate that the predicted binding site for 287 miR-760-3p is specific for this miR. MiR-34a-5p and miR-1307-3p were used as negative controls and, as 288 expected, neither miR binds to the 3'-UTR T100:TL duplex (S1 Fig. 2). Taken together, these experiments 289 show that host miR-760-3p binds to the SARS-CoV-2 genome 3'-UTR in its native fold, as we observe no 290 disruption of the original 3'-UTR T100:TL duplex mimic after binding with miR-760-3p.
. CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted May 19, 2023. ; https://doi.org/10.1101/2023.05.18.541401 doi: bioRxiv preprint Quantitative information about miR-760-3p binding to the 3'-UTR T100:TL duplex was obtained 292 by steady-state fluorescence spectroscopy. A modified 3'-UTR TL sequence containing a pyrollo-cytosine 293 tag (pyrC, excitation: 350 nm; emission: 445 nm) in the exposed bulge of the miR-760-3p binding site (5'-294 UCpyrCCCA-3', bolded in Table 1) was used in the preformation of the 3'-UTR T100:TL duplex mimic, significance by a Student's T-test reveals that the binding trends for miR-760-3p to the 3'-UTR T100:TL 305 duplex mimic and the full-length construct were not statistically significant from each other. In a control 306 experiment, we titrated miR-34a-5p to the pyrC-tagged 3'-UTR T100:TL duplex mimic, and observed no 307 quenching of fluorescent signal (S1 Fig. 3A). Additionally, pre-miR-125a was titrated to the DY547-miR- . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted May 19, 2023. ; https://doi.org/10.1101/2023.05.18.541401 doi: bioRxiv preprint 760-3p as a negative binding control for the full-length 3'-UTR, and again, we did not observe a significant 309 quenching of the signal (S1 Fig. 3B). This data, when taken together with our previous binding native 310 PAGE results, supports the specificity of the miR-760-3p binding interactions both to the 3'-UTR T100:TL 311 duplex as well as to the full-length 3'-UTR in its native fold.

312
Host miR-34a-5p and miR-34b-5p Bind Downstream of the SARS- 317 Thus, to mimic the miR-34a-5p and miR-34b-5p binding sites, we first used a sequence construct spanning 318 the entire s2m element (Fig. 1, green), its extended lower stem (Fig. 1, yellow), and the downstream hairpin 319 (Fig. 1, red), named here DIS-s2m extended (dimer initiation site-s2m extended). To assess binding 320 interactions between both miR-34a-5p and miR-34b-5p to the DIS-s2m extended, we performed native 321 PAGE experiments ( Fig. 4A and 4B). It has been reported that the s2m dimerizes in the presence of Mg 2+ 322 through the formation of a kissing dimer, which converts to an extended duplex structure, affecting its 323 migration in native PAGE experiments. Thus, we first analyzed the dimerization properties of the DIS-s2m 324 extended construct in comparison to the isolated s2m ( Fig. 1, green) and the DIS-s2m extended ( Fig. 1,   325 green, yellow, and red). Our results show that when incubated in the presence of increasing Mg 2+ 326 concentrations, the DIS-s2m extended shows a prominent dimer band (S1 Fig. 4, lanes 7-9), in contrast to 327 the isolated s2m which as reported previously forms a mixture of monomer, kissing dimer and extended 328 duplex conformations (S1 Fig. 4A, lanes 1-3). (6) The DIS-s2m extended dimer band could originate from 329 a mixture of three different conformations: kissing dimer, extended duplex and dimer stabilized through a 330 duplex formed by the extended hairpin region (Fig. 1, red) which does not involve the s2m initiated 331 dimerization (S1 Fig. 4C). To distinguish between these possibilities, we have used a DIS-s2m construct 332 which lacks the extended hairpin region (Fig. 1, green and yellow) and observed that this is primarily . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made concentrations for 24 hours, conditions in which we showed previously that promote formation of both 336 kissing dimer and extended duplex conformations (S1 Fig. 4B).(6) In these conditions, the DIS-s2m remains . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted May 19, 2023. ; https://doi.org/10.1101/2023.05.18.541401 doi: bioRxiv preprint mostly monomeric, but its dimer bands are more apparent (S1 Fig. 4B, lanes 4-6). Thus, we conclude that 338 in addition to dimer conformations mediated by the s2m, the prominent dimer band observed for DIS-s2m 339 extended also contains a dimer mediated through the extended hairpin, which is absent from DIS-s2m.

340
Since the nucleotides involved in miR-34a-5p and miR-34b-5p binding are located within the 341 predicted dimerization site involving the extended hairpin, we expected that the binding of either of these 342 miRs ( Fig. 4A and 4B, bottom), will disrupt this dimer structure of the DIS-s2m extended. As discussed

356
To confirm the assignment of the 1:1 complex as DIS-s2m extended with miR-34a-5p or miR-34b-357 5p complexes, we repeated the respective binding experiments using the Cy3 tagged miR-34a-5p and miR-358 34b-5p (excitation: 555 nm, emission: 563 nm) ( Fig. 4C and 4D). The Cy3-tagged miR-34a-5p and miR-359 34b-5p have a migration pattern change as compared to the unlabeled miRs due to the positive charge of 360 the Cy3 fluorophore. Using both Cy3-miR-34a-5p and Cy3-miR-34b-5p, we observed a strong fluorescent 361 signature that overlays with the previously identified complex band ( Fig. 4C and 4D, arrow 3), indicating 362 that this complex contains each of these miRNAs. However, we noted the appearance of a lower band that . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made 367 extended, proving that miR34a-5p and miR-34b-5p bind it specifically (S1 Fig. 5).

368
Next, we used steady-state fluorescence spectroscopy to determine the K d of both miR-34a-5p and 369 miR-34b-5p complexes with DIS-s2m extended, as well as to their complexes with the full-length 3'-UTR.

370
For binding to the DIS-s2m extended construct, we utilized a pyrC modified DIS-s2m extended, which has 371 the pyrC located within the 3'-tail of the DIS-s2m extended (5'-AGpyrCUGC-3', bolded in Table 1). We 372 performed experiments monitoring the binding of Cy3 tagged miRs to the full-length 3'-UTR, similar to 373 the DY547-miR-760-3p binding experiments. For the binding of miR-34a-5p to DIS-s2m extended, we 374 determined a K d of 11.7 ± 2.9 nM (R 2 = 0.9988) (Fig. 5A), whereas for its binding to the full-length 3'-375 UTR, we determined a K d of 21.3 ± 1.6 nM (R 2 = 0.9998) (Fig. 5B). Similar experiments were performed 376 for miR-34b-5p, where we determined a K d of 6.2 ± 2.2 nM (R 2 = 0.9975) to the DIS-s2m extended ( Fig.   377 5C), and a K d of 12.8 ± 1.3 nM (R 2 = 0.9997) for the full-length 3'-UTR (Fig. 5D). Analysis by a two-tailed 378 Student's T-test with equal variance revealed that for each miRNA, there is no statistically significant 379 difference between the binding curves to the DIS-s2m extended and the full-length 3'-UTR. Similar to the 380 native PAGE experiments, we utilized a negative binding control of miR-132-3p for the DIS-s2m extended 381 (S1 Fig. 6A) and pre-miR-125a for the full-length 3'-UTR (S1 Fig. 6B), and found no significant quenching 382 of the fluorescence intensity. These findings, when paired with our native PAGE analysis, indicate that 383 miR-34a-5p and miR-34b-5p both bind to the SARS-CoV-2 genome 3'-UTR with specificity to their 384 predicted binding site downstream of the s2m. was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted May 19, 2023. ; https://doi.org/10.1101/2023.05.18.541401 doi: bioRxiv preprint potentially used as a therapeutic option as a binding inhibitor to block miR binding on the SARS-CoV-2 394 genome 3'-UTR, relieving the wild-type miRs to perform their normal functions in the host.

395
We designed a FANA analog of miR-760-3p as a perfect complement to the 3'-UTR terminus 396 sequence, denoted as FANA-760 (Fig. 6A, top), as well as a FANA analog of miR-34a-5p and miR-34b-397 5p, denoted as FANA-34 (Fig. 6B,  to the 3'-UTR T100:TL duplex (Fig. 6A). FANA-760 exists primarily as a dimer (Fig. 6A,   . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  (Fig. 6B) revealed a highly similar binding pattern to that of both miR-34a-5p 410 and miR-34b-5p to the DIS-s2m extended in which binding of FANA-34 forms a 1:1 complex with DIS-411 s2m extended, as seen by the appearance of a band ~ 92 nt (Fig. 6B, bottom, arrow 3). This band appears 412 with a concomitant decrease in band intensity of both DIS-s2m extended dimer structures, which migrate 413 as a single band (Fig. 6B, bottom, arrow 4). Additionally, we noted that two bands appeared marking a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made fluorescence intensities in all competition assays were normalized to the first data point prior to titration of 433 the unlabeled ligand (FANA or miRNA) and fit with equation 2 to determine IC50 ( Fig. 7C and 7D) miR-34b-5p was determined to have an IC50 of 116.5  6.7 nM (Fig. 7D). From this information, a K I of 439 7.0  1.9 was calculated for FANA-34 to the full-length 3'-UTR construct (Fig. 7D). As a negative control 440 for the competition assay, we titrated miR-132-3p in place of the FANA-760 and FANA-34, and found no 441 increase of the fluorescence intensity (S1 Fig. 7), supporting the ability of the FANA oligomers to compete 442 with the wild-type miRNAs for binding to the full-length SARS-CoV-2 genome 3'-UTR and confirming 443 that the interactions are sequence specific.

Figure 7
. CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

445
In this work, we demonstrate that miR-760-3p, miR-34a-5p, and miR-34b-5p bind specifically and 446 with high affinity to the SARS-CoV-2 genome 3'-UTR. In several other viruses, miR "sponging," the 447 process of binding miRs to viral nucleic acids and/or proteins, has been directly correlated with virulence  was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted May 19, 2023. ; https://doi.org/10.1101/2023.05.18.541401 doi: bioRxiv preprint (lanes 4-9, arrow 5), followed by titration of either miR-34a-5p or miR-1307-3p, did not result in the 544 appearance of any higher molecular weight complex bands. No change in intensity of the 3'-UTR T100:TL 545 duplex band (lanes 4-9, arrow 5) was observed, supporting no binding between the 3'-UTR T100:TL duplex 546 mimic and the control miRs. (DIS-s2m extended) forms a dimer that is formed through the dimerization of this tail (lanes 7-9, arrow 7).

560
We also observe a kissing dimer (lanes 7-9, arrow 6) that is consistent with the isolated s2m dimerization

592
The K d used in this equation is the experimentally derived K d for each FANA or miR.
. CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted May 19, 2023. ; https://doi.org/10.1101/2023.05.18.541401 doi: bioRxiv preprint difference between K d values. (C) The K d for the miR-34b-5p:DIS-s2m extended complex was determined 645 to be 6.2 ± 2.2 nM, and (D) The K d for the miR-34b-5p-full-length 3'-UTR complex was determined to be 646 12.8 ± 1.3 nM. Student's T-test (two-tailed, equal variance) of the binding curve raw data points revealed 647 P (0.1561) > ɑ (0.01), indicating no statistically significant difference between K d values. . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted May 19, 2023. ; https://doi.org/10.1101/2023.05.18.541401 doi: bioRxiv preprint IC50 values of 56.1  2.1 nM for FANA-34, 117.4 ± 7.8 nM for miR-34a-5p, and 116.5 ± 6.7 nM for miR-670 34b-5p binding to the full-length 3'-UTR. . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted May 19, 2023. ; https://doi.org/10.1101/2023.05.18.541401 doi: bioRxiv preprint . CC-BY 4.0 International license available under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (which this version posted May 19, 2023. ; https://doi.org/10.1101/2023.05.18.541401 doi: bioRxiv preprint